The approximately 17,000 nursing homes in the United States provide care to over 1.6 million persons annually. Studies have indicated that between 60% and 90% of nursing home residents have diagnosable mental disorders (including dementia) (1,2,3). Despite the high prevalence of mental disorders among nursing home residents and their pressing need for adequate mental health care, nursing homes often lack the essential expertise, commitment, and financial incentives to ensure appropriate detection and treatment of mental disorders (4). The Omnibus Budget Reconciliation Act (OBRA) of 1987 required nursing homes to perform preadmission screening and annual resident review to ensure that persons with mental illness are not inappropriately admitted to nursing homes and that residents with mental illness who were appropriately placed receive necessary mental health services (5,6). Nevertheless, evidence suggests that in the early 1990s, after OBRA was implemented, the level of mental health care received by residents with mental disorders was still far below their actual needs (7,8,9,10). In a detailed chart review of over 2,000 Maryland nursing home residents in 1995, Fenton and colleagues (9) found that only 20% of residents had had a psychiatric consultation within 90 days of admission.

Enhancing the ability of and incentives for nursing homes to supply mental health services is a key component of steps to address this apparent discrepancy. Studies have suggested that on-site mental health services provided to nursing home residents resulted in improved clinical outcomes and reduced acute resource utilization (11,12). However, the function of the nursing home and its care and staff structures historically have focused on the management of chronic medical conditions and functional disabilities, rather than on psychiatric and behavioral abnormalities (8). As a result, expansion of nursing home service lines in response to the requirement of OBRA may take place slowly or barely take place, and it is expected to depend on intrinsic facility characteristics. Until now, however, no studies have described the longitudinal trends among nursing homes in their ability to provide mental health services during the post-OBRA era.

This study was designed to track the historical trend of the provision of mental health services in U.S. nursing homes, using the periodically conducted National Nursing Home Surveys between 1995 and 2004. Using the most recent data from 2004, this study further determined nursing home characteristics that may affect the ability to provide on-site mental health services and models of services.

Data sources

This study obtained the public-use, facility files of the 1995, 1997, 1999, and 2004 National Nursing Home Surveys (NNHS), which include cross-sectional and nationally representative samples of U.S. nursing homes collected by the Centers for Disease Control and Prevention (CDC). Details of survey methodologies and relevant reports from NNHS data can be found on the NNHS Web site (13). Briefly, the sampling frame of each year's NNHS consisted of all nursing homes that were certified by Medicare or Medicaid and nursing homes that were not certified but licensed by individual states. In each year the survey involved a stratified two-stage probability design where the selection of nursing homes in the first stage was stratified by bed size, certification status, and other facility characteristics; nursing homes were finally selected with the probability proportional to bed size. The probability sampling, together with rigorous data collection protocols and data editing mechanisms, ensures that each survey contained a nationally representative sample of nursing homes, including their services, staff, and residents (14,15,16,17).

In each survey, specifically trained field interviewers collected facility data through personal interviews with nursing home administrators. Completed questionnaires were entered into a computer database by data specialists from the CDC, and extensive editing was conducted to ensure that all responses were accurate, logical, and complete. The number of nursing homes included in the sample was 1,409 in the 1995 NNHS, 1,406 in the 1997 NNHS, 1,423 in the 1999 NNHS, and 1,174 in the 2004 NNHS, with the annual response rate ranging between 81% and 95%.

Variables

The NNHS collected a useful set of nursing home characteristics each year, and the latest survey of 2004 contained more comprehensive information. Each survey asked whether mental health services were provided in the nursing home (the outcome variable of interest). Other facility variables collected in all surveys included whether the nursing home was chain affiliated, the profit status of the facility (for profit or nonprofit), bed size (categorized as <50, 50–99, 100–199, and ≥200), and whether it was located in a metropolitan statistical area (MSA). For nursing homes that provided on-site mental health services, the 2004 survey also asked whether the services were available regularly or at routinely scheduled times (regular basis), on an on-call basis (or as needed), or on both a regular and on-call basis. Additional facility variables collected in the 2004 survey were percentage of Medicare residents (categorized as <10%, 10%–19%, and ≥20%); percentage of Medicaid residents (<20%, 20%–39%, 40%–59%, 60%–79%, and ≥80%); whether the nursing home was accredited by the Joint Commission on Accreditation of Healthcare Organizations, Commission on Accreditation of Rehabilitation Facilities, or Continuing Care Accreditation Commission; and census region (Northeast, Midwest, South, and West).

Statistical analyses

The study first performed univariate analyses of each cross-sectional sample to summarize nursing home characteristics. Stratified analyses were further performed to track over time the percentage of nursing homes that provided on-site mental health services, according to whether they were a for-profit or nonprofit facility, whether they were affiliated with a chain, bed-size category, and whether the facility was located in an MSA. Because the CDC did not release sample weights for the 1995 facility data, which would have allowed for population extrapolation, statistics for these analyses represented sample estimates. Sensitivity analyses were performed that took into account the complex sampling design of the data in years other than 1995. Given the large sample of facilities in each year and the fact that the sampling process likely affected the estimated standard errors but not point estimates, the results in the sensitivity analyses (that is, cross-sectional estimates and trend over time) remained similar and therefore are not reported here.

Multivariate analyses were subsequently performed focusing on the 2004 data. First, the study estimated a binary logistic regression model of on-site availability (yes or no) of mental health services as a function of the intrinsic nursing home characteristics described above—bed size collapsed into three categories (<100, 100–199, and ≥200) and percentage of Medicaid residents (≥20% or <20%). Additionally, a multinomial logistic regression model was estimated to determine the independent associations of nursing home characteristics with the likelihood of having a particular model of providing mental health care—on a regular basis, on an on-call basis, or on both a regular and on-call basis. Nursing homes with each model of service provision were compared with the group of nursing homes that did not provide on-site mental health services.

In order to account for the complex sampling methodologies in 2004, including stratification, clustering, and weighting, both regression models were estimated by using the survey estimation routines contained in Stata, version 8.0, special edition. The study reported odds ratios (ORs), 95% confidence intervals (CIs), and p values estimated from the regression. This funded project was approved by the institutional review board of the University of California, Irvine.

Trends over time

The percentage of nursing homes that provided on-site mental health services to their residents was 75% in 1995, 81% in 1997, 80% in 1999, and 78% in 2004 (Table 1). In 2004, 25% of all nursing homes provided mental health services on a regular basis, 24% on an on-call basis, and 28% on both a regular and on-call basis.

Table 1 also shows that between 1995 and 2004, 54% to 60% of nursing homes were chain affiliated, over 30% were nonprofit, and over 60% were located in an MSA. Although 21% of nursing homes had ≥200 beds in 1995, this percentage declined to less than 10% thereafter. Figure 1 shows that over the period of 1995–2004, for-profit facilities were more likely than nonprofit facilities to supply on-site mental health serves, and chain-affiliated facilities were more likely than non-chain-affiliated facilities to provide mental health services, although group differences were not substantial in either case (that is, <10%) and the longitudinal trend for each type of facility was flat.

On the other hand, larger facilities and facilities in an MSA were more likely than their counterparts to provide on-site mental health services (Figure 1). When facilities with ≥200 beds were compared with facilities with <50 beds, the analysis found that the differences were 21% in 1995 (85% versus 64%), 14% in 1997 (91% versus 77%), 19% in 1999 (86% versus 67%), and 25% in 2004 (95% versus 70%). Compared with non-MSA facilities, MSA facilities had an increased rate of providing on-site mental health services: the difference was 11% in 1995 (78% versus 67%), 7% in 1997 (84% versus 77%), 8% in 1999 (83% versus 75%), and 21% in 2004 (84% versus 63%). The increased difference in 2004 resulted mainly from the reduced proportion of non-MSA facilities providing mental health services from 1999 to 2004.

Characteristics associated with provision of services

In 2004 several facility characteristics were found to be independently associated with the on-site availability of mental health services and models of services. Table 2 shows that on-site mental health services were more likely to be available for larger facilities (≥200 beds) than for smaller facilities (<100 beds) (OR=3.80, p=.024), for facilities having 10%–19% of Medicare residents than for those having <10% of Medicare residents (OR=1.53, p=.015), and for MSA facilities than for non-MSA facilities (OR=2.60, p<.001). Compared with facilities in the Northeast, facilities in other regions had substantially reduced service capacities (ORs ranged from .17 to.27; p≤.001 for all cases except for the Midwest, where the p value was .001).

In the analyses of available service provision models (Table 3), bed size, proportion of residents receiving Medicare or Medicaid, census region, and MSA status were still found to be significant predictors. Facilities with ≥200 beds were over four times as likely as facilities with <100 beds to provide mental health services on a regular basis or on both a regular and an on-call basis; facilities with ≥20% of Medicaid residents were more than three times as likely as other facilities to provide mental health services on a regular basis or on both a regular and on-call basis; and facilities in regions other than the Northeast were 12% to 38% as likely as facilities in the Northeast to provide any of the three service provision models.

This study based on nationally representative surveys of U.S. nursing homes revealed that after OBRA was implemented, between 1995 and 2004 approximately 80% of nursing homes provided on-site mental health services. Overall service availability in nursing homes did not change over time, and facilities in nonmetropolitan areas showed a decline in providing mental health services between 1999 and 2004. Facility characteristics including bed size, Medicare and Medicaid census, and location in an MSA were positively associated with on-site provision of mental health services, and location in regions other than the Northeast predicted a substantial reduction in both overall service availability and specific models of service provision (that is, on a regular basis, on an on-call basis, or on both a regular and on-call basis.

Although OBRA was intended to provide new protections for nursing home residents and persons with mental disorders, analyses in this study suggest that more than 3,000 nursing homes in the nation (that is, 22% of all nursing homes) did not seem to be able to serve appropriately persons with mental illness. In addition, as shown by the 2004 data, expansion in nursing home care to serve the unique needs of persons with mental illness tended to be more difficult for smaller facilities, those without a larger proportion of Medicaid or Medicare residents, and those in rural areas. In addition to the stipulations contained in OBRA, a number of relevant regulatory requirements, reimbursement policies, and quality assurance efforts (discussed below) were initiated during the study period. However, the relatively stable numbers over time suggest that these initiatives may have had little impact on the overall access to mental health care for nursing home residents.

Indeed, policies and initiatives over this period may be inconsistent and create disincentives for nursing homes to supply on-site mental health care. For example, during the initial implementation stage of OBRA in the early 1990s, Medicare benefits were expanded to cover mental health services delivered in nursing homes by clinical psychologists and social workers. However, Medicare expenditures for nursing home-based mental health services tripled in subsequent years, and concerns arose that residents may receive unjustified treatments that are not beneficial to those with mental disorders (18). As a result, as of 1999 Medicare suspended independent reimbursement to nursing homes for mental health services provided by social workers (19). In addition, the Balanced Budget Act of 1997 repealed federal standards for state Medicaid programs to reimburse nursing home care and allowed individual states to set their own rates. The overall low rate of Medicaid payment, and particularly the much lower rates set by states experiencing budget cuts (20,21), may not ensure that residents with mental illness receive appropriate nursing home care or access to specialized care.

According to federal mandates, nursing homes are subject to annual scrutiny by state surveyors. These surveyors use on-site inspections, interviews with representative nursing staff and residents, and selected medical record reviews to identify deficiencies in care, including mental health care, and to mandate corrections and possible sanctions for identified deficiencies (22). However, widespread concerns exist about whether state inspections can ensure appropriate nursing home care (23,24), and weaknesses of state oversights on mental health services in nursing homes have been repeatedly reported by the federal Office of the Inspector General (OIG) (25,26). In responses to these OIG reports, the Centers for Medicare and Medicaid recently called for an improved survey and certification process targeted at residents with mental illness (27). However, the impact of this and other efforts of state inspections on ensuring better access to mental health care is yet to be evaluated.

Medicare and many private insurance programs do not currently provide for parity in mental health care coverage. For example, Medicare-covered psychosocial services in nursing homes require a 50% copayment rather than the regular 20% copayment for medical and surgical care, which tends to impede nursing home residents' access to essential behavioral treatments. The facts that private insurers may have more restrictive plan policies and that Medicaid provides supplemental coverage for out-of-pocket expenses may explain the findings that nursing homes catering to a higher percentage of publicly insured residents were more likely to provide on-site mental health services.

Nevertheless, the overall inadequate coverage for mental health care in nursing homes by Medicare and Medicaid programs, which together account for the majority of nursing home payments, discourages specialty mental health providers from serving persons with mental illness. For example, a survey of approximately 900 nursing directors in six states found that the availability of psychiatric consultations in nursing homes was deemed to be inadequate by half of nursing directors and that nonpharmacologic management and behavioral interventions, which tended to be uncovered or poorly reimbursed, were largely overlooked by consulting specialists (8). As a further note, the gross shortfall of geriatric psychiatrists and other professionals (28) tends to worsen the substantial unmet needs for specialist care by residents with mental illness, especially those residing in small or rural facilities and facilities in certain regions (8).

Further analyses of service models in this study suggested that in 2004, 25% of nursing homes provided mental health services on a regular basis, 28% on both a regular and on-call basis, and 24% on an on-call basis, with the remaining 22% of facilities providing no on-site mental health services. This suggests that the ability to ensure adequate mental health services is also likely to vary among facilities that do provide such services. Bartels and colleagues (12) reviewed relevant literature on models of providing on-site mental health services in nursing homes and found that the traditional consultation-liaison service on a one-time, on-call basis tended to be least effective in improving clinical outcomes and reducing acute care utilization, such as hospitalizations. They concluded that the routine presence of mental health specialists combined with a multidisciplinary team is a preferred model of delivering mental health care in nursing homes. Nevertheless, it is often the case that nursing homes cannot often arrange formal contracts with consultant psychiatrists to ensure their routine availability (8). Therefore, additional efforts are needed to remove the financial, institutional, and system-level barriers to obtaining routine specialist care for nursing home residents with mental illness.

This study has several limitations. First, although the analyses were based on well-conducted, nationally representative surveys of U.S. nursing homes, the CDC did not release more complete information on nursing home characteristics in the public-use data, which prevented fuller analyses of both the historical trend and cross-sectional correlates of service availability. Research is needed to investigate other facility, geographic, and policy factors that affect the provision of appropriate mental health services in nursing homes. Second, nonresponding nursing homes in the survey may have biased the results. However, given the large sample and high response rate in each survey, response bias may not be an important issue.

A third limitation is that although this study analyzed the availability of on-site mental health services and, when available, models of provision of mental health services in nursing homes, detailed information was not available on the actual care received by residents with mental illness and its adequacy and quality. The public-use NNHS data do not include unique facility and resident identifiers that would allow us to identify residents and evaluate their actual needs for mental health care and actual care received. However, given the high prevalence of mental illnesses among nursing home residents (1,2,3), the finding of this study that a large number of nursing homes did not provide appropriate on-site mental health services suggests a significant policy and practice issue. Finally, the analyses spanned the decade of 1995–2004 and found that the overall availability of nursing home-based mental health services did not improve over this period. Further studies are needed to determine whether it is still the case in more recent years.

This study suggests the consistent inability of more than 3,000 nursing homes in the nation to provide on-site mental health services, which corroborates earlier evidence that a substantial number of residents with mental illness did not have adequate access to needed services during the post-OBRA era (7,8,9,10). The findings also suggest that the lack of access may be more pronounced for smaller or rural facilities, facilities in regions other than the Northeast, and facilities serving a small number of publicly insured residents. It is imperative that governments, regulators, policy makers, and health care professionals work together to craft supportive mechanisms that can expand the availability of essential mental health services in nursing homes.

The author gratefully acknowledges funding from R01AG032264 from the National Institute on Aging.

The author reports no competing interests.